Nrurtng
Elsevier
HRR
49
Reueurch, 32 (1988) 49-64
01021
Representation
of amplitude modulation in the auditory cortex of the cat.
II. Comparison between cortical fields
Christoph
Mas-Plunck-Instrrur~r
E. Schreiner
hiophwkalrsche
(Received
15 August
* and John V, Urbas
Chrmte, Ahrr~lung Nrurohiolo~i:ie, C;iirtitl,yrn. E R. t;.
1986; accepted
27 October
1947)
The responses of neuronal clusters to amplitude-modulated
tones were studied in five auditory cortical fields of the ancsthetiLed
cat: the primary auditory field (Al). second auditory field (AII), anterior auditory field (AAF). posterior auditory field (PAF) and the
ventro-posterior
auditory field (VPAF). Modulation
transfer functions (MTFs) for amplitude-modulated
tones were obtained at 172
cortical locations. MTFs were constructed
by measuring firing rate (rate-MTFs)
and response s~nchr~nizatioll
(synchr~~nizati(~n-MTFs)
to sinusoidal and rectangular
waveform modulation
of CF-tones, The MTFs were characterized
by their .best-rn[~duiati(~n frequency’
(Q z,ie). These characteristics
(BMF) and a measure of their quality or ‘sharpness’
were compared
for the five fields. Rate and
synchronization
MTFs for sinusoidal and rectangular
modulation
produced
similar estimates of BMF and Qz,,t~, Comparison
of
averaged BMFs between the cortical fields revealed relatively high BMFs in AAF (mean: 31.1 Hz for synchronization
to sinusoidal
AM) and moderately
high BMFs in AI (14.2 Hz) whereas BMFs encountered
in All. VPAF and PAF were generally low (7.0, 5.2.
and 6.8 Hz). The MTFs were relatively broadly tuned (low Qzda) in AAF and sharper in a low modulation
group containing
AII.
PAF and VPAF. The ventro-posterior
field was the most sensitive to changes m the modujati(~n waveform. We conclude that there
are significant
differences
between auditory cortical fields with respect to their temporal
response characteristics
and that the
assessment of these response characteristics
reveals important
aspects of the functional significance of auditory cortical fields for the
coding and representation
of complex sounds
Auditory
cortex:
Amplitude
modulation:
Cat
Introduction
In the cat. at least seven different
auditory
cortical fields have been described
(Reale and
Imig, 1980). An important
parameter
by which
these fields are physiologically
distinguished
is
their cochleotopic
organization,
i.e. the spatial
orientation
of the frequency gradient derived from
the neuronal best frequencies.
Other functional differences between fields have
been described for sharpness of tuning (‘spectral
resolution’},
latencies,
minimum
thresholds,
and
Cbrrrspcvrde~rc~e to: John V. Urbas, Max-Planck-Institut
fir
biophysikahsche
Chemie.
Abteilung
Neurobiologie.
3400
G~itting~n. F.R.G.
* Currmr u&‘rc.~~: Coleman Laboratory,
School of Medicine.
HSE 863. Department
of Otolaryngology,
University
of
C‘ahfornia, San Francisco, CA 94143-0526. U.S.A.
the spatial distribution
of neurons with different
binaural
response
properties
(e.g. Phillips
and
Irvine, 1980. 1981; Imig and Adrian, 1977: Middiebrooks et al., 1980; Phillips and Orman. 1984;
Schreinet
and Cynader,
1984). However.
these
parameters are insufficient
to distinguish
functionally between some auditory
fields. F‘or example.
the primary auditory
cortical field (Al) and the
anterior
auditory
field (AAF)
receive similar
anatomical
projections
and have very similar response thresholds,
latencies and frequency tuning
(Merzenich et al., 1975: Knight, 1977; Phillips and
Irvine, 1980, 1981; Imig and Reale, 1980: Reale
and Imig, 1980; Andersen
et al.. 1980: Imig and
Morel, 1983). So far, the only known functional
parameter clearly distinguishing
these fields is the
orientation
of their gradient of characteristic
frequencies
(CFs), i.e. by properties
of the main
parameter in the spectral domain.
Other
signal
characteristics
that could
he
utilized to distinguish
between functional
properties of cortical fields are parameters
in the tcmporal domain. The temporal patterning
or modulation of frequency and amplitude over time is an
important
information-carrying
feature of complex sounds such as species-specific
communications sounds.
Complex
communication
signals
contain modulation
frequencies
over a range extending from a few events per second to up to
several hundred Hertz. In this report we attempt
to throw some light on the role of the auditory
cortex in the coding and representation
of temporally varying signals and to document
differences
in basic temporal response properties of neurons
among five cortical fields: anterior auditory field
(AAF), primary auditory field (AI) (including
the
‘dorsal zone’ (Middlebrooks
and Zook. 1983)).
second auditory field (AII). posterior auditory field
(PAF) and ventro-posterior
field (VPAF).
Regular amplitude
changes of a carrier signal
(amplitude
modulation
(AM)) were used for this
exploratory
study of temporal processing
within
these auditory cortical fields. Temporal
‘modulation transfer functions’ (MTFs) of multi-unit
responses have been obtained
for the parametric
description
of the findings. Results are limited to
those properties that have the potential to identif!
major interfield differences. General properties of
cortical responses to AM signals which appeared
to be common to most fields (e.g. level dependence. binaural
interaction
etc.) have been described for AAF in the first part of thih studv
(Schreiner and Urbas. 1986).
Materials and Methods
The experimental
procedures utilized have been
described in more detail in the first part of this
study (Schreiner
and Urbas, 1986). Experiments
were performed on 15 adult cats. Initially.
ketamine hydrochloride
(30 mg/kg i.m.) was administered to allow venous cannulation.
An initial intravenous
dose of pentobarbital
(Nemhutal
5
mg/kg) and subcutaneous
injections
of atropine
(1 mg) and a corticosteroid
(Solu-Decortin-H.
Merck. 10 mg) were then administered
to establish
a surgical level of anesthesia
and to control the
secretion
of mucus and development
of brain
edema. respectively.
Followmg
tracheal cannul;rtion and placement
in a head-holder.
the cat wah
paralyzed
with gallamine
triethiodidc
(8 mg,/ kg
i.v.) and artificially
respirated.
Throughout
the
experiment
the animal
was infused
with 5
mg/kg/h
gallamine
triethiodide
and 2 mg/kg/h
Nembutal in a solution of 209 glucose III Ringer.5
solution (infusion volume: 3 ml/h). End-tidal CO,
and ECG were continuously
recorded and maintained within stable limits with appropriate
measures, e.g. supplements
of the anesthctlc. ‘The temperature was maintained
at 37.-38”<‘.
After exposure of the lateral cortical surface. a
well was constructed
and filled with mineral oil.
Recordings
were made from the left hemisphere
with low impedance
(approximately
1 MS2 at 1
kHz) carbon fiber microelectrodes.
Peristimulustime histograms (PSTHs) were constructed
for the
responses from small groups of neurons (2 --5 cells).
and occasionally
from single units. The recordings
reported here were derived at an intracortical
depth
ranging
from about 800- 1200 microns. roughly
corresponding
to cortical layers III and IV. To
allow the comparison
of different cortical fields.
recording of multi-unit
activity was restricted to
layers III and IV which produce the largest and
most reliable responses. To minimize the relative
contribution
of thalamo-cortical
axons to the rc’sponse. activities with latrncies shorter than X-Y
ms were not included into the data because of the
strong probability
that they did not originate from
cortical neurons. Cortical fields were identified by
their locations relative to the sulcal pattern. by the
gradients of the encountered
CFs. and especially
for AII, by the sharpness
or quality of derived
tuning curves.
Auditory
stimuli
were presented
via headphones (Sennheiser
HD424) positioned directly at
the pinnae and calibrated in situ with a Briiel and
Kjaer
sound-level
meter.
The
characteristic
frequency (CF) threshold,
Q,,J,lH (C‘F divided bj
bandwidth
of tuning curve 10 dB above minimum
threshold) and binaural response characteristic
for
neurons and neuronal
clusters at each recording
site were determined.
This was accomplished
b>
using 200 ms tone bursts and examination
of the
stimulus-synchronized
spike record on the oscilloscope and by listening to the amplifier output fed
into a loudspeaker.
Amplitude
modulation
(AM)
51
signals were then presented
to the contra-lateral
ear. These stimuli employed carrier frequencies at
the CF that were amplitude-modulated
(100%
modulation
depth) by either sinusoidal
or square
wave envelopes.
Amplitude
modulation
stimuli were delivered
as bursts of 1 s duration
alternating
with silent
intervals of 0.8-1.5 s duration. PST histograms of
the response to AM stimuli were derived at every
site for a range of modulation
frequencies
(14
frequencies
between 2.2 and 200 Hzj. The signal
level of each MTF was held constant
at a level
lo-40
dB above minimum
response
threshold.
PSTH binwidth
was 1 ms; usually, 20 response
averages were collected for each stimulus setting.
Two kinds of temporal
modulation
transfer
functions (MTFs) were constructed
from the PST
histograms of responses to AM tones. ‘Rate-MTFs’
were derived from the mean firing rate of a group
of neurons during the presentation
of a modulated
tone and plotted as a function of the modulation
frequency. This rate-MTF
was normalized
to the
maximal mean firing rate for any modulation
at
that location.
‘Synchronization-MTFs’
were derived from the synchronization
(or phase-locking)
of the neuronal response to the modulation
of the
carrier signal. Synchronization
is a measure of the
accuracy of the coupling between the neural response and the temporal properties of the stimulus. The hamming-weighted
PSTH was Fourieranalyzed and the contribution
of the first plus the
second harmonic
of the modulation
frequency
(relative to the dc-component)
were taken as a
synchronization
metric. The second harmonic was
included
into the analysis
to permit
an appropriate
assessment
of responses
that showed
on-responses
as well as pronounced
off-responses
to a single period of the AM, thus resulting
in a ‘frequency
doubling’
of stimulus-triggered
events in the histograms.
Strong off-responses
were seen mostly
for rectangular
AM. The
‘synchronization-MTFs’
were constructed
relative
to the maximal synchronization
at a given site and
plotted as a function of the modulation
frequency.
Both MTFs will be described in this report by
two parameters.
The first is the modulation
frequency that produces the maximum synchronization (or maximum mean firing rate) in the neuronal
response,
termed
the ‘best modulation
frequency’
(BMF). The second parameter
is a
measure of the quality of the MTF. In this study,
the bandwidth
2 dB below the maximum
(approximately
20%’ below the tip of the MTF) will
be used to quantify
the quality:
‘QZdD’ = BMF
divided by bandwidth
of MTF?,,
below BMF.
More details of the procedures and representative
examples of individual
synchronization
and rate
MTFs are given in the first part of this study
(Schreiner and Urbas. 1986).
Results
Best modulation frequency for gwchronization
For sinusoidal
stimulation,
synchronization
MTFs have been obtained
at 172 sites in five
auditory
cortical
fields.
A best
modulation
frequency (BMF) could be determined
at 88% of
these locations, covering a modulation
range from
3 Hz---l00 Hz. The other 12% showed a low-pass
characteristic
with a maximum at the lowest modulation frequency used in this study (2.2 Hz).
The distributions
and mean values of the BMFs
for the synchronization
of the response
to
sinusoidally
modulated
CF tones encountered
in
AAF. AI, AII, PAF. and VPAF are shown in Fig.
1. Although
the distribution
of the encountered
BMF values was ragged and possibly bimodal (e.g.
Fig. 1A. D) mean and standard
deviation
have
been calculated
to facilitate comparison
between
the properties of different cortical fields. Considerable differences of mean and standard deviation
between these fields were obtained.
The highest
average BMF was found in AAF (average 31.1 i
27.4 Hz). The average BMFs in the other fields
were 14.2 + 8.8 Hz in AI, 7.0 f 5.0 in AII. 5.2 5 3.2
Hz in VPAF. and 6.8 h 11.2 Hz in PAF. Analysis
of variance
was significant
(F-test.
P < 0.05).
Conservative
statistical evaluation
(post-hoc comparison, Method of Scheffe) demonstrated
a significant difference (P -c0.05)
of the average BMF
of AAF from the BMFs in AI. AII, PAF and
VPAF. In addition,
the BMF of AI was significantly different from those of AII. PAF and VPAF,
whereas the BMFs of AII, PAF and VPAF showed
no significant
differences.
Consequently.
the investigated cortical fields can be loosely grouped
into three categories with regard to their ‘temporal
resolution’ (or ability to follow temporal changes
tion field’, and a group of ‘low resolution I’ields’
AII, PAF and VPAF.
Rectangular modulation of a carrier signal produces response characteristics similar to sinusoidal
modulation.
However. rectangular modulation
produced a higher incidence of off-responses than
sinusoidal modulation. At least in AI, there was
some indication that the locations with strong
off-responses were spatially clustered indicating
that the off-responses may be a component of the
‘ temporal processing mechanisms’.
In Fig. 2A the (averaged) BMFs for sinusoidal
and rectangular modulation are shown for the five
contained
_
/i
AI
sync
j / sinus
/
N
-
C,,T--....:,j y”,
...i
,
”
!
I
BMF
---zq
A
32. r
AI1 j
sync./
r-----.----
.-------
_synchronization
slnus.
I
-
E
BMF
$8 T-.-------
AAF
A I
A II
’ PAF
‘VPAF
sync
6
stnus
15
Y
g
:
05
1
t
o-T-~
Fig. 1. Distribution
of best-moddation
frequencies (BMFsf of
the modulation-transfer
functions (MTF) for the synchronization of responses to sinusoidal amplitude modulation
of a CF
tone. The distributions
are shown for five auditory
cortical
fields. Arithmetic mean and standard deviation of the distributions are indicated.
Each bin contains BMFs up to but not
including the next higher frequency marked on the abscissa.
as expressed by the average BMF for the synchronization to sinusoidal AM): AAF represents a
‘high resolution field’, AI was a ‘medium resolu-
..T- i
AAF
nr:
~.
A I -- A II -_ PAF- ’ VPAF
Fig. 2. (A) Averaged BMFs of the response synchronization
to
sinusoidal (open bars) and rectangular
(shaded bars) amplitude
modulation.
Error bars represent
the standard
error of the
mean. The standard
deviations (SDS) for the sinusoidal modulation
are given in Fig. 1. The SDS for the rectangular
modulation
were 13.7 (AAF), 6.5 (Al), 4.4 (AH), 4.8 (PAF). 3.9
(VPAF). (B) Synchronization
index of responses to AM. The
averaged maximum synchronization
in each of the investigated
cortical fields was determined for the first and second harmonic
of the modulation
frequency
and added (maximal
possible
synchronization
index: 2). The SDS for the fields were 0.39
(AAF), 0.33 (Al), 0.42 (AU). 0.23 (PAF), and 0.18 (VPAF).
53
cortical fields. The differences and si~larities
between temporal
characteristics
of the fields are
equally expressed for rectangular
and sinusoidal
AM, although the BMFs in AAF and AI tended
to be of slightly lower frequencies for rectangular
AM. The synchronization-BMFs
of AAF for rectangular modulation
were significantly
different (P
< 0.01) from the other four cortical fields. With a
P < 0.1, the average BMF of AI exhibited only a
statistical trend to be different from the fields AIL
PAF and VPAF. The differences between the later
fields were not statistically
significant.
In general, the ability of auditory cortical neurons to follow amplitude-modulated
(or repetitive)
signals showed only minor influences
from the
shape of the modulation
envelope. No significant
influences of modulation
waveform on the average
syn~hrollization-BMF
of a field were found for
AI, AIL PAF and VPAF. Only AAF showed a
slightly higher synchronization-BMF
for sinusoidal
AM as compared
to rectangular
AM (P -z 0.05;
see Schreiner
and Urbas,
1986). A response
parameter more strongly influenced
by the stimulus shape was the amplitude
of the synchronization and rate MTFs (see below), i.e. the degree of
synchronization
and responsiveness,
respectively.
The low temporal resolution of fields AII, PAF
and VPAF was coupled with a smaller degree of
synchronization.
or phase locking to the modulation, when compared to the fields with medium or
high temporal resolution (AI, AAF). This is indicated in Fig. 2B in which the mean of the synchronization index obtained at BMF for a given field
is plotted. AAF and AI showed, on the average,
an anlount
of synchronization
that was higher
than the synchronization
in AII, PAF and VPAF
(P c 0.05). The maximal synchronization
in VPAF
was lower (Y K 0.05) than that of AAF, AI and
AII. The relative influence of stimulus waveform
on the degree of synchronization
is described below.
The relationship
of synchronization-BMFs
with
the CF of the corresponding
site is shown in the
scatter diagram in Fig. 3. The sampled CF range
covers approximately
l-40 kHz in AAF, AI1 and
PAF. whereas the range of AI and VPAF was
limited to sites with CFs above 2-4 kHz. For
AAF a significant
correlation
between CF and
BMF was observed, with higher BMFs recorded at
high CFs and lower BMFs seen for lower CFs
(Y = 0.63. P < 0.01; see Schreiner
and Urbas,
1986). Besides AAF. no other field showed a
significant correlation
between CF and BMF.
BMFs were not highly correlated with the quality or sharpness
of tuning curves. Only AAF
showed a small but significant
correlation
(r =
0.32. P <: 0.05) between
BMF and QIOdtf (see
Schreiner
and Urbas. 1986). The QIOdn of the
multiunit
recordings (derived at most locations in
this study) appeared
to be slightly lower than
comparable
results derived from single unit studies (Philhps and Irvine, 1980, 1981: Merzenich et
al., 1975).
Best modulation frequency for firing mte
Synchronization
is a measure of the accuracy of
the coupling between the neural response and the
temporal
properties
of the stimuius.
Firing rate
provides a different measure of the neuronal
responsiveness
to a stimulus. In peripheral auditory
stations,
firing rate and synchronization
can be
largely independent
measures of the response activity. The influence of amplitude
modulation
on
the firing rate in different cortical fields will be
described in this section.
The shape and general feature of rate-MTFs
obtained
in the auditory
cortex were similar to
synchronization-MTFs
recorded at the same site.
In Fig. 4, the distribution
of the BMFs of rateMTFs of the five cortical fields is summarized.
They reflect a relatively
broad distribution
for
AAF and AI and somewhat
narrower
distributions for AII, PAF, and VPAF. The means of the
five distributions
show a tendency similar to the
synchronization-MTFs
with the highest average
BMF (27.0 t- 26.1 Hz) obtained
for AAF,
a
medium high BMF for AI (19.9 t_ 26 Hz), and low
BMFs fOJ AI1 (6.6 + 3.8 Hz), VPAF (7.1 + 6.0
Hz), and PAF (8.2 i 16.0 Hz). Fig. 5 shows the
averaged rate-BMFs
for sinusoidal
and rectangular AM. both reflecting
interfield
differences
of
similar magnitude.
Statistical analysis (Method of
Scheffe) of the rate-BMFs
for sinusoidal and rectangular modulation
led to a similar categorization
scheme of the temporal
resolution
as seen for
synchronization-BMFs.
Again, the five fields can
be grouped into high, medium and low temporal
resolution
fields. The high resolution
field AAF
B
.;;_ 4 AAF
‘11
.
5_-
sInusoIda
synchron
Y:5
.
‘ I.
4
;
.. .
.
..a..
.
.
.
,^. -
.. .
. .
” = 39
Bf+
. .
.
.
51
.
--
--
.
- N=4
l
. .
r-
cc+, A I
1.z
1 smusoldal
SCY synchron.
. .
-
BMF
I
2c,
“=I9
l
.
.
.
.
l
I’
‘f,_.___
.--.
,.
.-.
.
,;I
r.r.7
-
CF
0
3
IOG-$AF.-LZ
.
.
-0
.
5d
i
-
*.*
7
_I,._
0205
.
.
5
-
E
.
.
.
’
r’----‘2
'
.
I
T
? .
LZ
.
-v-..’
20
10
.
:, _
m
l
.
15
.
l
.
3-
.
.. ..
.
..
.
’ sinusoidal
53 ti synchron
30
kH2
CF
-.
_..--.
_
l@DW-..
HZ
smusoldal
synchron
50
A N-2
BMF ,,j
t
I
n=25
lC-
.
-
.
.*
5 :
no
"..
222
y--,
02
'T--,---T
05
7
2
1
-
.
l .m
.
1
I
”
.
3
Fig.
a.
l
!
,
5
17
-r
10
.
T--T'
20
I
30
.'
kuz
CF
3. Scatter diagram
of BMF as function
of CF for 5 auditory
cortical fields. The BMFs were determined
from
synchronization-MTFs
for sinusoidal AM. Number of animals (N) and number of sites in each field (n) are indicated
has significantly
higher BMFs (P < 0.05)than all
other fields. The medium resolution
field AI has
significantly
lower BMFs than AAF (P -C0.05)
the
and exhibits a statistical trend toward higher BMFs
than the fields in the low resolution
group (P -c
0.1). The low resolution group contains the fields
55
A '11.
B
n ~34
125
8
AAF
rate
slnus
i
6
t
1
i
2
0i
__
=199
"
=261
SC=26C
AI
rote
sinus
13
8
6i
d
23L
5 7 1c"520%0751001
-
D
Ti
12
113
N
,L
ir
-+
BMF
li,
n=27
12
PAF
E
1
BMF
-
BMF
1f.T--12
10
N
rate
slnus
8
6
L
2
(
0
I
,
,
TJ
I
/
r-y-l
o 2 3 L 5 7~o'520305d5~oo
0
HZ
-BMF
-
BMF
Fig. 4. Distribution of BMFs of rate-MTFs
for sinusoidal AM in five auditory cortical fields. Mean and standard
deviation
indicated. Each bin contains BMFs up to but not including the next higher frequency marked on the abscissa.
AIL PAF and VPAF which showed no statistically
significant
differences
among their average rateBMFs.
Differences
between
rate-BMFs
for
--
p&
AAF
Fig.
bars)
The
SDS
Fig.
AI
cl
slnusoldol AM
&j
rectangularAM
A II
PAF
VPAF
5. Averaged BMFs of rate-MTFs
for sinusoidal
(open
and rectangular
(shaded bars) amplitude
modulation.
error bars represent the standard error of the mean. The
of the BMF distributions
for sinusoidal AM are given in
4. The SDS for rectangular
modulation
were 22.1 (AAF),
16.9 (AI). 3.5 (AU), 1.5 (PAF), and 3.9 (VPAF).
are
sinusoidal or rectangular
modulation
were statistically not significant.
A scatter
diagram
of the rate-BMFs
for
sinusoidal modulation
plotted as a function of CF
is shown for each field in Fig. 6. Again, only AAF
showed a significant
correlation
between
BMF
and CF (r = 0.67, P < 0.01). In the other fields
the BMF appears to be independent
of the CF
although the evidence in fields AI and VPAF is
incomplete
because of the somewhat limited CF
range of the sample.
Bundwidth of cortical MTFs
In addition
to the BMF, another
parameter
describing
the MTFs is the sharpness or quality
factor of the transfer function. The bandwidth
2
dB below (20% down from) the tip of a MTF was
used to calculate
the quality factor of the AM
transfer characteristic
(BMF divided by the bandwidth 2 dB below BMF). As illustrated
in Fig. 7.
the distribution
of Qzda in most cortical fields was
between 0.5 and 4.0 for the synchronization
to
sinusoidal
AM. The widest (least selective) syn-
B
A
, ! Aif
..
smsoidal
rate
s.7
:!
.
c.
!
.
.
I
! N-4
BMF
20.~
.
a
00
l *
‘(I..
.
1
I
.
0.0
;
.
,
LT..’ ‘7--T r--J
00
:f)
. ._....
T_‘..
r;
_-
.-
25
__.j
1
:;
i
-T---2c
3c.
-77-T
.c
.6L$2’d5”-
.'I
D r-
CF
_..~._
c r.'--_.-~.
~.__.._-.._-_ ~.___
-
2
-.
:co--PAF
20
,
.
.
.
.
.
*
.
-..
,
.
05
r
,
I
,
1
2
.
--_----
E ---
/
5
n=27
12 -
.
.
.
l
3:
.
.
.
.
l
a
.m.
.
em
a.'
22>J
.”
T?T-T-J
,
10
CF
.
.
.
.
I
7
N=L
5-
“*
.
,
!
I
.
l
l
. .
3-r
02
.
1
NH2
CF
.“.:
.
16
t
. .
l
BMF
3c
,
-.-_.
’ slnusotdal
50-I
rate
N=3
n=30
.
'0
~2
l
.
.
n=26
.
. .
.
.
.
.
.
_ slnusoldal
VII-! rate
.
.
.*
.
.
l*
.
.1
-42
..
.
.
.
.
.
I ;
t
..
.
N=5
n =34
.‘::_.
.
BMF
I,,; A
.
20
30
k-7
__
_-..
20 30
_
kHZ
CF
100-j VPAF
bZ
BMF
t
-! slnusoldal
so - rate
; N=2
,,A
n=22
;
:Q -4
-00
.
1
5
3
22,
1
a
*+--+
..;
”
.
00
20
-
.
0
30
kH?
CF
Fig. 6. Scatter diagram of BMF as function of CF of rate-MTFs for sinusoidal modulation in five auditory cortical fields.
chronization-MTFs
were found in AAF. All other
studied cortical fields had on the average significantly narrower synchronization-MTFs
(P < 0.05)
of about equal Q2dB (Fig. 8A) for sinusoidal modulation. The QZdB for synchronization-MTFs
derived from rectangular
modulation
indicated
a
57
A32
,x
2-I
-129
n =37
SDz066
N
t
1
;;=$$
-Q
Q,,,
-n
q
2
3
i
5
2dB
26~1
VPAF’
sync
slnus
Fig. 7. Distrtbution
of QzdB of synchronization-MTFs.
QrdR is a measure of the sharpness or quality of the modulation-transfer
function (BMF divided by the bandwidth
of the MTF 2dB - approximately
20% - below the tip of the MTF). Mean and standard
deviation are indicated for each cortical field.
statistical trend for AAF and AI of smaller temporal selectivity
than found in AII, PAF and
VPAF.
A
The QZdB of the rate-MTFs
of different fields
(Fig. 8B) showed no differences
for sinusoidal
waveforms.
However. for rectangular
modulation
B
37
I
_synchronlzat~on~
r:
stnusoldal
[.
rectangularAM
AAF
A I
3
rate
AM
A II
PAF
VPAF
AAF
A I
A II
PAF
VPAF
Fig. 8. (A) Averaged Qlde of synchronization-MTFs
for sinusoidal (open bars) and rectangular
(shaded bars) amplitude modulation.
Standard deviations were (sinusoidal,
rectangular):
0.66, 0.87 (AAF); 1.02, 0.70 (AI); 1.02, 0.99 (AII); 0.90. I.11 (PAF); 1.03. 0.91
(VPAP). (B) Averaged Qzde of rate-MTFs for sinusoidal (open bars) and rectangular
(shaded bars) amplitude modulation.
SDS were
(sinusoidal. rectangular):
1.05, 0.63 (AAF); 1.20. 0.90 (AI); 0.96. 1.03 (AII); 1.00, 1.18 (PAF); 0.82, 0.95 (VPAF).
significant
differences
(P -c 0.05) were seen between AAF and the low resolution
group (AII,
PAF. VPAF). In addition, the &,a of AI showed
a statistical
trend towards being lower than the
values for PAF and VPAF.
Corresponding
to the categorization
scheme for
the temporal resolution
based on the BMF, the
‘temporal
selectivity’ of the MTFs based on the
2dB
measure
can be used as an additional scheme
Q
in the attempt to functionally
group cortical areas.
A ‘low selectivity field’ (AAF) and a ‘high selectivity group’ with AII, PAF and VPAF could be
distinguished.
AI occupied an intermediate
position between the former fields with no unequivocal statistical
evidence
to establish
a separate
‘ medium selectivity field’.
Stimulus shape
As noted above, it appears that the shape of the
modulation
waveform, in this case sinusoidal
or
rectangular,
had only a small influence
on the
BMF at a given cortical location.
However, the
waveform did have some influence on the shape of
the MTFs, or, more specifically, on the width and
height of the MTFs. By quantitatively
comparing
the influence of sinusoidal versus rectangular
AM
on responsiveness
(firing rate) and degree of synchronization,
it can be shown that there are field
characteristic
properties
in the signal processing
that are sensitive to the shape of a stimulus. Fig. 9
shows the ratio of the response
strength
of
sinusoidal versus rectangular
stimuli averaged for
I
05
0
AAF
A I
AI1
PAF
VPAF
Fig. 9. Averaged ‘shape ratios’ of synchronization
(open bars)
and rate (shaded bars). ‘Shape ratio’ is defined as the ratio of
maximal firing rate - or synchronization
- of sinusoidal MTFs
to rectangular
MTFs. The average was determined
for CFs
above 5 kHz since some fields have not been sampled sufficiently below 5 kHz. Standard
deviations
(synchronization,
rate) were: 0.31, 0.35 (AAF); 0.2, 0.35 (AI); 0.35, 0.51 (AII);
0.22, 0.55 (PAF); 0.19, 0.51 (VPAF).
each field. These ‘shape ratios’ were calculated by
dividing the absolute response strength at the peak
of a sinusoidal MTF by the strength at the peak of
the rectangular
MTF at the same site. Shape ratios
for rate and synchronization
MTFs are shown. On
the average, all shape ratios for synchronizationMTFs had values below 1. This means that, on the
average, rectangular
AM produced a higher synchronization
of the response to the stimulus envelope than sinusoidal AM. The lowest shape ratios
for synchronization-MTFs
were found in AI (0.42
& 0.36) and VPAF (0.46 k 0.23). For both fields
these ratios were significantly
lower than the ratios
in the other fields (P -C0.05). This indicates
a
relatively strong influence
of the stimulus waveform on the synchronization
particularly
for AI
and VPAF.
Two different
aspects can be noted for the
influence of the modulation
waveform on the firing rate. Firstly, there are two fields (PAF and
VPAF) that had shape ratios with values equal or
larger than 1, indicating
equal or greater responsiveness
of these fields to sinusoidal modulation compared
to rectangular
modulation.
The
shape ratios of these two fields were significantly
higher (P < 0.05) than ratios obtained
for other
cortical fields. This finding is in sharp contrast to
the usually observed tendency of neurons to respond more vigorously or more precisely to rectangular modulation
than to sinusoidal modulation
as indicated
by the rate profiles of AAF, AI and
AI1 and the synchronization
indices for all cortical
fields (see Fig. 9).
Secondly, the shape ratios for rate and synchronization-MTFs
significantly
differed for PAF and
VPAF (P -c 0.05). Thus, it appears that the posterior and in particular
the ventro-posterior
field
may be able to differentiate
between modulation
waveforms and reflect these differences by means
of their response mode. Rectangular
modulation
produced a relatively high synchronization
whereas
sinusoidal
modulation
produced a relatively high
firing rate. The other cortical fields did not show
significant influences on their response mode (e.g.
rate versus synchronization)
that could easily be
related to changes in the waveform.
Sinusoidal
and rectangular
modulation
waveforms were selected for these experiments
because
they represent two extremes with regard to their
59
spectral and temporal characteristics.
It should be
noted that using other ‘shapes’ than those as
stimulus
envelope
(e.g. asymmetrical
triangular
modulations
may possibly lead to stronger effects
on the responsiveness
to AM sounds than shown
here for sinusoidal and rectangular
AM. However,
the results described above indicate that a ‘shape
sensitivity’ of the response does not strongly influence the best modulation
frequency
of the
transfer function,
the main parameter
describing
temporal response characteristic.
Discussion
Ternporul resolution
In the five auditory cortical fields studied, the
neuronal
response properties
to amplitude
modulated signals were characterized
by features of
the obtained modulation-transfer
functions,
most
notably
the best-modulation
frequency
and the
quality
factor QZdB. It appears
that responses
within each cortical field were ‘tuned’ to a range
of modulation
frequencies.
Within
each of the
investigated
fields, these frequency
ranges were
similar for different aspects of the response properties as indicated
by similar BMF ranges for
synchronizationand rate-MTFs.
This result indicates that the synchronization
and rate measures
are not largely independent
response features at
the level of the cortex. Due to the phasic response
character
(response
confined
to the onset of a
stimulus) of the majority of the responses, firing
rate and timing of the response to each AM cycle
was closely linked. Thus, the high correfation
between timing and magnitude
of the response can
be expected.
Differences in BMF range, due to variation of
the stimulus waveform,
were small in AI, AII,
PAF and VPAF. Only for AAF, significant
differences in the response range of the MTF characteristics could be seen between sinusoidal
and
rectangular
modulation
(Schreiner
and Urbas,
1986). In AAF, the BMFs for rectangular
modulation were slightly lower than for sinusoidal
AM,
suggesting a relatively strong influence of the rise
time and/or the spectral spread of the modulation
waveform on the MTF.
The relative
independence
of the temporal
characterization
of the recorded activity from response measures
(i.e. synchronization
vs. firing
rate) and from stimulus
aspects (i.e. rectangular
vs. sinusoidal modulation)
suggests that these temporal characteristics
represent a fundamental
feature of cortical processing.
This consistency
provides a strong argument
for the validity of the
modulation
analysis as an additional
descriptor
for functional
characteristics
of a given cortical
field.
An important
finding of this study is that some
cortical fields respond on the average to slightly
different
modulation
frequency
ranges. In other
words, some auditory cortical fields may be statistically distinguishabIe
by their overall ability to
resolve in their response certain temporal variations of the stimulus amplitude.
The distribution
of best-modulation
frequencies
of various locations within a cortical field provides an approximation for an overall estimate of the ‘temporal
resolution’ of a given field. Since there is considerable overlap of the BMF distribution
of different
cortical fields, the BMF at a given cortical location by itself does not allow an identification
of
the corresponding
cortical field. With the exception of AAF, the width of the BMF distribution
of
the cortical fields is fairly narrow and appears to
be largely independent
of the CF location within a
field. In AAF, on the other hand, a correlation
between BMF and CF location was verified, explaining
the relatively
broad BMF distribution
found for that field. In comparison,
the CF range
for the locations
in AAF was the least biased
toward high CFs. Consequently.
the average BMF
values for AAF were slightly lower than the results of a CF specific average. Despite this slight
bias toward smaller differences between the average BMF of AAF and the other fields, the AAF
values were found statistically
significantly
different from temporal response characteristics
of the
other fields, justifying
the use of average AAF
values instead of a CF specific analysis.
Some
indications
for possible bimodal distributions
of
BMFs (e.g. in PAF) suggest a higher complexity of
the relationship
between
BMF and other functional parameters than this exploratory study could
resolve and reveal.
Based on their average BMF range, auditory
cortical fields were grouped with regard to their
‘temporal
resolution’:
1) a ‘low temporal resolution group’, containing
AII. PAF and VPAF. with
a preferred
BMF range of about 667 Hz: 2) a
‘medium
resolution
field’. AI. with a temporal
resolution about twice as high as the former group;
and 3) a ‘high resolution field’. AAF, with another
increase of the temporal resolution by a factor of
approximately
two.
As discussed
in the first part of this study
(Schreiner and Urbas. 1986) an influence of the
anesthesia
on the observed temporal characteristics cannot be excluded. However. the field differences with regard to temporal response properties found under similar anesthetic conditions
and
the lack of evidence that the temporal resolution
in unanesthetized
animals largely differs from in
the anesthetized
animal
(Whitfield
and Evans.
1965; Evans, 1974; Creutzfeldt et al., 1980; Saitoh
et al.. 1981) suggest that the described results give
a fair. although sketchy, picture of some temporal
processing properties of the auditory cortex.
Other concerns with the interpretation
of multiunit data from the middle layers of the auditory
cortex are a) whether the recordings
represent a
functionally
homogeneous
population
and b) what
is the amount of contamination
of the recording
with responses originating
from thalamo-cortical
axons. Activities with latencies shorter than 8-9
ms were not included in the data, thus eliminating
a portion of the activity most likely not originating from cortical neurons (see also Schreiner and
Urbas. 1986). However, neither the homogeneity
of the recorded population
nor the contribution
of
thalamo-cortical
axons can be verified with the
applied methods. Consequently.
the given interpretations of the obtained data in terms of cortical
processing
have to be considered
with the appropriate caution.
Temporal selectivity and ‘shape sensifivit):’
A second feature characterizing
the temporal
response properties of cortical fields is the sharpness or quality of the obtained
MTFs, e.g. expressed through Qzde. This MTF feature can be
considered a measure of the ‘temporal selectivity’
of a field: the narrower or sharper the MTF, the
more selective is the response to a certain temporal range of signal variations.
Three ranges of
temporal selectivity of cortical fields emerge. The
relatively weakly defined ranges of high. medium
and low temporal
selectivity correspond
well to
the more sharply
defined
field characteristics
found for temporal resolution with the effect that
fields with a high temporal resolution tend to have
a low temporal selectivity and vice versa. Fig. IO
illustrates the relation between the main temporal
characteristics
for fields with high.
response
medium and low temporal resolution. The average
mean and standard deviation of BMF and QZ‘rR is
shown as it appears when all four investigated
aspects. i.e. synchronization.
rate. sinusoidal
and
rectangular
modulation.
arc combined.
For the
group with low temporal resolution. the data from
AII, PAF and VPAF were pooled for clarity. The
close correspondence
of interfield differences determined
by different
aspects of the MTFs emphasizes again the usefulness of amplitude
modulated signals for a more general functional
classification of cortical fields.
Three of the five investigated
cortical fields
were relatively
insensitive
to variations
of the
modulation
waveform,
indicating
that the repetition rate is the dominant
factor determining
the
temporal resolution
(de Ribaupierre
et al.. 1972:
Creutzfeldt
et al., 1980). However, a large proportion of responses
in PAF and VPAF did show
some sensitivity with regard to the stimulus waveform. This observation
suggests the existence of
some spectral/temporal
processing
capacities
in
those fields that go beyond a simple repetition
rate concept. This ‘shape sensitivity’ is reminiscent
of some aspects of the ‘shape selectivity’ of cortical units to noise bursts, as described by Swarbrick
and Whitfield (1972) and Bock (1976).
Temporal resolution and /uncrronul specifk7f.r~
The temporal resolution
of the auditory cortex
stands at the end of a processing sequence along
the ascending auditory pathway, that progressively
diminishes
the ability to consistently
follow fast
fluctuations
of the signal envelope. This phenomenon has been discussed
in connection
with the
anterior auditory
field in the first paper of this
series (Schreiner
and
Urbas.
1986: see also
Schreiner and Langner,
1988). In general, modulation rates in the input signal that are preserved
61
i
i
2
Q-2dB
Fig. 10. Average range of temporal resolution and temporal selectivity for three categories of auditor? cortical fields. The average
to sinusoidal and rectangular
AM, and
means and standard deviations of BMFs and Q&a were derived from synchronization-MTFs
from firing rate-MTFs
produced by sinusoidal and rectangular
AM. For the group of cortical fields with low temporal resolution
(AII, PAF and VPAF) the values of all three fields were combined.
in the temporal activity pattern of neuronal
elements decline from up to 1000 Hz in the auditory
nerve to a few cycles per second in the auditory
cortex. This progressive
change in the temporal
response properties
may reflect consequences
of
processing
strategies
that extract, combine
and
‘information-bearing
paramedistribute
certain
ters’ (Suga, 1978, 1984). The difference
in the
temporal properties of the auditory cortical fields
suggests that these temporal changes are not exclusively
a consequence
of a loss of temporal
precision, e.g. due to summation
of synaptic transmission jitter.
The apparent sequential processing involved in
the reduction
of temporal
resolution
along the
auditory pathway up to the cortex can not be used
for the explanation
of the temporal disparity found
between some cortical fields. Anatomical
studies
have clearly shown that the thalamic input to the
cortical fields is organized in parallel projections
(Andersen
et al., 1980: Imig and Morel, 1983).
Each of these projections
contains a unique combination
of contributions
from different thalamic
nuclei. Little is known about the temporal properties of these thalamic
nuclei. Rouiller
and colleagues (1981, 1982) studied the response properties of different
parts of the medial geniculate
body (MGB) using click stimulation.
They found
only a few units that appeared to be tuned to a
certain repetition rate of the clicks. Of these units.
the ‘best repetition
rate’ showed only minor differences between various nuclei. Most of the ‘best
repetition
rates’ in the MGB were found to be
between 50 and 70 Hz, far above the BMF range
of most cortical locations.
Pars lateralis and pars
magnocellularis
of the MGB showed some differences in the sharpness of the temporal resolution defined by click trains (Rouiller et al.. 1981).
Creutzfeldt
et al. (1980) demonstrated
a dramatic
reduction in the temporal resolution between synaptically connected
neurons of the MGB and the
auditory cortex of unanesthetized
guinea pigs. This
finding, in connection
with the highly branched
projection
pattern of afferents originating
in the
MGB (Andersen
et al., 1980; Imig and Morel,
1983) may suggest that the temporal characteristics of each cortical field are produced
in each
field itself. i.e. it is most likely a result of a parallel
processing scheme.
Ablation
studies of the auditory
cortex (Diamond and Neff, 1957; Kelly. 1974; Neff et al.,
1975) demonstrated
the involvement
of cortical
processing in the identification
of tonal sequences
or patterns.
The temporal
structure
of those sequences and patterns
whose recognition
was affected
by the
ablation
was
in a similar
frequency/repetition
range as the temporal resolution of the auditory cortex as demonstrated
with
AM signals. This indicates that the temporal features of the auditory cortex obtained with IMTFs
can indeed he of behavioral
significance.
Limitations on the nature of the ‘information-bearing
parameters’
represented
in the cortex of the cat
arc imposed by the spectral and temporal
filter
characteristics
of a given unit as expressed
b>
frequency
tuning curve and modulation
transfer
function.
The largely independent
assessment
of
spectral and temporal filtering characteristics
of
cortical neuronal elements can be used to account
for a large portion of the actual response characteristics to more complex signals (Schreiner
et
al.. 1983; Schreiner and Langner. 1987) and therefore represents
a first step in defining
in more
detail specific properties of auditory cortical fields.
This exploratory
study of the representation
of
amplitude
modulation
in different cortical fields
of the cat provides only a limited view of temporal
processing
characteristics
and capacities
of the
auditory cortex. Already in this first. still crude.
assessment
some intriguing
features of cortical
temporal processing principles could be discerned.
although this study represents only a small step in
defining
the complex
structure
of the signal
processing network of the auditory cortex. More
studies, in particular with emphasis on single unit
recordings,
are necessary
concentrating
on the
parametric representation
of potentially
important
signal features in distinct cortical fields.
Acknowledgements
Diamond.
I. 1. and Ncff.
and discrimination
t:\ans.
E.F. (1974)
tic pattern
Program.
prtcch\sh
for sound
F.O. Schmitt
Study
ences. Third
pp. 131
Imig.
Neural
and
211
for the detcc[lon of ;~cc,u\-
locali~.;ctmn. In.
and F.(i.
Program.
I hlrd
t F.&L
W’orden
(‘nmhridge.
Srudt
Neurobcl-
MA\: \!tT
I’re\\.
145.
T.J. and .Adrian.
primary
H.O.
field (AI)
(lY77)
BInaural
of cat auditor>
columns
corvz.
1n IIIC‘
BraIn Re\
13X.
241. 2.57.
Imig. T.J. and Morel.
Y5
A. (lY82)
Orgamc.atmn
hystem in the CAL .Ann.
cat auditory
ot lhalamocor[lRev
Ncuroscl.
6.
120.
Imig. T.J. and Reale.
connections
cortex. J. Camp.
Kelly.
J.B. (1974)
temporal
R.A.
(1980)
Pattern\
to
tonotoplc
Neurot.
192. 293
related
Potysensorq
pattern
of cor&o-corucat
map5
In cat
auditor)
332.
cortlcat
discriminarion
lesions .md auditor)
in rhe cat. Bram Res. 80.
317.. 327.
Knighl.
P.A.
anIer]or
(1977)
Represenlarion
auditory
field (AAF)
of
the wxhtea
In the
of the caf. BraIn
Res. 130.
Evidence for neuronal penodrclly
derecnon
447.467.
l.angncr.
G. (1983)
in [he auditory
system of the gumea fowl: Implications
pitch
in the time
analysis
domain.
Exp.
Brain
for
Res. 52.
333-355.
Merrenich.
M.M..
\cntatlon
Knight.
P.L. and Roth. G.1.. (tY75)
of the cochlea in the primary
Ihe cat. J. Neurophysiot.
Middlebrooks.
J.M.
(1983)
uon of the cat’s medial geniculate
to
Jecrmn\
auditor?
binaural
J.H..
cortex of
Intrmslc
orgamza-
hody identified
response-specific
cortex. J. Neurnsci.
Middlehrooks.
3. 203-224.
Dykes. R.W.
and Merzenich.
response specific bands in primary
the
Topographical
W.D..
organization
I.T.
and Casseday.
loural htudics of auditorv
and W.D.
Neff
(Eds.).
Vol 5,/Z. Springer.
Phillips.
D.P.
D.R.F.
Behav-
In: W.D.
Keidet
of Sensory Physiology.
hit>. J. Neurophysiol.
primary
audirory
rumng and response\
cor[cx
LC,Inten-
45. 4X-.5X.
and Irvine,
neurons in the anterior
(IYXI ) Response> of smgle
defined
(?rI) of the car: frequency
D.P.
4X.
Berlin.
and Irvine.
neurons in physiologically
Phillips.
of
10 ISO-
J.H. (1975)
discrimmation.
Handbook
.M.M. (198Ot
cortex (Al)
orthogonal
I X1.31.
Brain Rcs.
Diamond.
by pro-
bands in rhe prlmar)
BInaural
UI
Reprc-
auditop
3X. 231-249.
J.C. and &ok.
frequency contours.
References
.Ahlation 01 rcmporal COTIC\
parlern. J. Ncur<q~h\\I,‘t
300 -?l 5.
Neff.
We thank Prof. Dr. 0. Creutzfeldt
for his support of this study and Dr. T. Lewien for providing
computational
facilities.
We also express
our
thanks to Dr. T. Allard for his advice with regard
to the statistical analysis and for making numerous stylistic suggestions regarding an earlier draft.
( lY57)
D.
of auditor\
D.R.F.
(1982)
auditory
Propertrea
field (AAF)
of bmgle
of cal cerebral
cortex. Brain Res. 24X. 237.-244.
Andersen,
R.A.. Knight.
rhalamocorticat
P.L. and Merzenich.
and corticothalamic
and the anterior
auditory
.M.M. (1980) The
connections
field (AAF)
of AI. AI1
in the cat: evidence
for two largely segregated systems of connections.
Neural.
Bock. G.R.
10
194, 663-701.
(1976)
0..
Thalamocortical
Hellweg.
D.P.
cortical
neurons
noise. J. Acoust. Sot. Am. 59. S47.
F.-C.
transformation
and
Schreinrr.
of responses
audirory stimuli. Exp. Brain Res. 39. X7-104.
c‘.
10
(1980)
complex
and
Orman,
ncuronh in posterior
stimulalion.
Rcale.
R.A.
auditory
Responses of cat auditory
amplitude-modulated
Crcutzfeldt.
J. Camp.
Phillips,
Ribaupierre.
(1972)
S.S. (1984)
J. Neurophysiot.
and
Imig,
Response
field of cat auditory
T.J.
single
10 tonal
51. 147.163.
(19X0)
Tonotopic
cortex of the cat. J. Camp.
F. de, Goldstein.
of
cortex
M.H.
Neurol.
orgamzation
192. 265-291.
and Yeni-Komlshian.
C‘orhcal coding of repetitive
G.
acoustic pulses. Brain
Rcs. 4X. 205-225.
Rouitter.
E. and Ribaupierre.
F. de (1 Y82) Neurons
brnsitlvc 10
63
narrow ranges of repetitive acoustic transients in the medial
geniculate body of the cat. Exp. Brain Res. 48, 323-326.
Rouiller,
E., Ribaupierre,
Y. de, Toros-Morel,
A. and
Ribaupierre.
F. de (1981) Neural coding of repetitive clicks
in the medial geniculate body of cat. Hear. Res. 5. 81-100.
Saitoh. K., Maruyama,
N. and Kudoh. M. (1981) Sustained
response of auditory cortex units in the cat. In: Y. Katsuki,
R. Norgren
and M. Sato (Fds.). Brain Mechanisms
of
Sensation. John Wiley and Son. New York.
Schreiner. C.E. and Cynader. MS. (1984) Basic organization
of
second auditory cortical field (AII) of the cat. J. Neurophysiol 51. 128441305.
Schreiner,
C.E. and Langner.
G. (1984) Representation
of
periodicity in the inferior colliculus of the cat. Sot. Neurosci. Abstr. 10. 34.5.
Schreiner.
C.E. and Langner. G. (1987) coding of temporal
pattern\ in the central auditory nervous system. In: G.M.
tdelman.
S. Hassler and W.M. Gall (Eds.), Functions
of
the auditory system, J. Wiley and Sons, New York.
Schreiner. C. and Urbas. J.V. (1983) Functional
differentiation
of cat auditory cortex areas demonstrated
using amplitude
modulation
(AM). Neurosci. Lett. Suppl. 14, S334.
Schreiner, C.E. and Urbas. J.V. (1986) Representation
of amplitude modulation
in the auditory cortex of the cat. I. The
anterior auditory field (AAF). Hear. Res. 21. 2277241.
Schreiner. C.. Urbas. J.V. and Mehrgardt,
S. (1983) Temporal
resolution of amplitude modulation
and complex signals in
the auditory cortex of the cat. In: Hearing
Physiological
Bases and Psychophysics.
edited by R. Klinke and R.
Hartmann.
New York: Springer-Verlag.
Suga, N. (197X) Specialization
of auditory system for reception
and processing of species-specific
sounds. Federation
Proc.
37. 2342-2354.
Suga.
N. (1984)
Neural
mechanisms
of complex-sound
processing for echolocation.
Trends Neurosci. 6. 20-27.
Swarbrick. L. and Whitfield. I.C. (1972) Auditory cortical units
selectively
responding
to stimulus
shape.
J. Physiol.
(London) 224. P68-69.
Whitfield. I.C. and Evans, E.F. (1965) Responses of auditory
cortical neurons to stimuli of changing frequency. J. Neurophysiol. 2X. 655-612.